TwitterLinked in

LTi ASPIRE banner

Long Read: foiling application fraud

20/11/2026

Share/Bookmark

Identity fraudContributing Editor Allan Foad writes: This article can be read as a standalone piece, or as an annex to my contribution to last month’s Leasing World Long Read under the title, Are you sure you know who you are dealing with?

Both articles deliver the message that the only way to counter fraud generated by artificial intelligence is with artificial intelligence, and today I’m focusing on application fraud.

It is often claimed that application fraud is the neglected child in the fraud family because there are much bigger and more prevalent frauds for businesses to combat. Statistics show that out of the nearly 34 million applications submitted last year, only 150,000 proved to be fraudulent representing 0.44 percent of the total. In truth automated vetting and authorisation systems are actually very good at detecting fraud in applications submitted singularly. Still, that 0.44 percent has a big impact to those affected.

There is a way a way to counter these fake AI generated applications but before I expand on it, I would like to segue into describing a real-life case study that will prove pertinent later.

A friend of mine recently hired a local building company to undertake major works at his house. Shortly after appointing the company, which is privately owned by one man, my friend received an email saying that the company had changed its banking arrangements, and you can guess the rest. My friend actually knows the builder very well and was immediately wary, but he was 99.9pct convinced the email was a fake when he saw that it was sent out at half past two. in the morning. He knew there was no way the builder would have been in his office or working from home at that time.

Looking at the time zones, he reasoned that it might have originated somewhere in Eastern Asia, possibly China, or it could have been sent out as a package of emails by a robot (more commonly known as a bot). Wherever it came from, it was not from the builder.

I received this email as well because I am a trustee of a charity that he is also working for. Both were signed using the builder’s usual cursive script which made them look even more authentic, but a more forensic investigation revealed that whilst the builder’s genuine email address is set out in lower case, the letters of his firm’s initials in the fraudulent version were in upper case.

When we spoke to the builder he realised he had been hacked and emails had been sent out to his full client list. When enquiries were made, he discovered that at least one client had thought it was genuine and had acted upon it. Fortunately, the client’s bank refunded the money and the builder is not out of pocket.

Returning to the main thrust of this article, the reason why a substantial number of the successful frauds beat the system is because the finance company is not aware that multiple applications have been sent out, hoping that some of them will convince someone somewhere, as happened in the anecdote above. 

The most graphic example of this happening was during Covid when the government made “bounce back” loans available to small and medium sized companies to enable then to survive and recover. It was a laudable initiative and these were vital to many enterprises. What the government did not realise was that fraudsters were submitting multiple applications and walking off with the money, significantly contributing to the £11 billion lost during this period. 

This remains an embarrassment to the previous government, but it could have been avoided if officials had not assessed each application in a silo. If they had stored all of them in a database and searched for common trends looking for multiple applications with the same IP address or replicated wording for example, they could have been alerted to fraudulent activity and much of this money would have been saved.

The government has said it plans to establish an Online Crime Centre in collaboration with the regulators, banks and finance companies, telecommunication firms, and police with a view to establishing a database of all financial applications. That database will then be searched by AI tools to unearth multiple applications and other common trends, to pick out warning signs. Maybe it will look out for applications arriving in the middle of the night! The objective is for credit providers to search this database before processing an application and hopefully be alerted to fraudulent ones before they are processed. This should make life harder for the fraudsters.

I think the concept of this initiative is impressive on paper but am mindful that this database is going to be a very rich store of information that will be a target for fraudsters to break into [in terms of risk vs reward, this is surely a more attractive target for fraudsters than dialling up the temperature controls on fridges]. On top of that, how does this square with the rules on client confidentiality and the Data Protection Act? It is all a big ask but that said, I can see how it could prevent fraud if it is allowed to go ahead.

There is no doubt that AI has been a boon to the major fraudsters who are operating all over the world, and there is a perception that they have mastered its capabilities faster than the authorities. It is also pretty clear that there is a need to fight fire with fire and for good AI to be pitched against the bad.

I am glad to be a spectator in this fight!

  

* This is a shortened and updated version of the article which first appeared in Leasing World magazine, issue 221.


Comment on this article

Share/Bookmark


Back